The present invention relates generally to the conversion of a digital audio signal into an analog audio signal with pre-amplification, and in particular, to a microprocessor controlled, single-bit Pulse Density Modulated (PDM) digital-to-analog converter (DAC) incorporating digital and analog preamplifier stages capable of processing digital audio data formats toward the generation of low-noise analog audio signals. Currently, most digital audio playback devices incorporate single chip, multi-bit digital-to-analog converter technology and provide no internal pre-amplification of the audio signal.
Prior to the advent of digital audio technology, where "live" audio performances are recorded digitally for later playback, sound recording was performed exclusively in the analog domain generating various formats such as LP albums. Such methods utilized a recording process which resulted in the impression of a "physical copy" of the analog signal into a vinyl disk creating a "groove" which was "tracked" by a phonograph needle the physical movement of which would reproduce an electrical "copy" of the groove. This method of sound reproduction, however, suffered from a number of problems including a low signal-to-noise ratio (SNR) and audible distortion. Digital technology sought to overcome these problems by converting the analog audio signal, at its source, into a digital format. The conversion of analog audio signals which are by definition continuous in nature into digital audio signals which are discreet in nature is performed by a process known as quantization which involves taking a "snapshot" of the analog audio signal a various points in time and transforming the voltage at that time to its corresponding digital word, utilizing one of the standard digital audio formats. The process of taking snapshots is otherwise known as sampling and the frequency of sampling is referred to as the sampling rate.
Currently, there are a number of standard formats of encoded digital audio data including: AES/EBU, EIAJ CP-340, and S/PDIF. Most digital technology currently uses a 16-bit digital word format which is the result of the encoding process and which are later used to reproduce or "recover" the original analog sound. Accurate reproduction of the original analog signal requires that the digital sampling to occur at least twice the rate of the highest frequency to be digitized. If the sampling rate is in excess of twice the highest frequency, then it produces greater accuracy in the measurement of the analog signal. This approach to quantization is known as oversampling.
At present, compact audio disks are a common media for "storing" digital audio recordings where the digital audio data is encoded serially on the disk. A typical compact disc player optically reads the digital data off the encoded disc; converts that data using a single chip, multi-bit digital-to-analog converter to an analog audio signal; and presents this signal to various output plugs and pins (RCA, headphones, coaxial) for connection to analog audio components such as preamplifiers or power amplifiers. Higher-end digital audio components in addition to providing analog audio outputs also provide digital audio outputs which contain the digital audio data prior to conversion usually on optical and/or electrical output connectors.
One shortcoming of some current digital audio apparatus is the use of higher order filters to control the signal-to-noise ratio in the outputted analog signal. These higher order filters tend to cause both amplitude and phase aberrations due to the need for cascading. While some compact disc players attempt to compensate for these problems through the use of oversampling, which tends to ease requirements on post conversion filters, these solutions often fall far short of the audio performance that high end audio users desire.
Another limitation in some current compact disc/digital audio designs is the use of the traditional multi-bit digital-to-analog converter devices. The problem with multi-bit digital-to-analog converters is the difficulty balancing the multiple outputs to create a linear signal This design defect can be overcome by precise balancing of each separate circuit; however, this solution is difficult, expensive and unworkable due to further variations in each bit processor caused by temperature changes within the unit.
A further limitation of single-chip digital audio equipment is the introduction of noise through clock jitter and the failure to isolate audio circuits from noise-producing sources such as oscillators. While isolation is desirable, it is impossible to achieve when a single chip component is used since they often contain in an integrated format the various devices which one desires to separate from one another. Jitter reduction presents a greater problem that can be reduced by minimizing the path length between circuits which produce the clock signals and circuits which use the clock signals. However, minimizing the path length may place the oscillator near the audio circuits, thus reintroducing noise, creating a "catch 22" situation.
Accordingly, the present invention seeks to address the foregoing limitations of the prior art digital audio equipment by providing an improved digital-analog pre-amp and converter apparatus which comprises a microprocessor controlled single bit Pulse Density Modulated digital-to-analog converter with hybrid pre-amplification in both digital and analog domains toward improving the recovery of analog audio signals from the digital format with less distortion and lower SNR.
It is an associated object of the present invention to provide means of up-sampling the digital audio data by 128 times the incoming rate through the use of a finite impulse response filter and delta-sigma modulation resulting in noise-shaping converter.
Another object of the present invention is to provide for the utilization of an 18-bit digital audio format, in place of the common 16-bit format thereby allowing for higher dynamic range; obviating the need to truncate data; and providing lower quantization error.
As rule of thumb in digital audio technology, it is considered that each additional bit in a digital audio word contributes 6 dB to system performance, thus the addition of 2 bits from the traditional 16-bit digital audio word to the 18-bit word utilized in the present invention provides an additional 12 dB of dynamic range amounting to over a 10% increase from the prior art. The increased word size in the present invention provides for future digital audio formats which will tend to have longer word lengths. Currently, the AES/EBU format provides for the possibility of 24-bit digital words.
Handling a 24-bit word in a 16-bit digital-to-analog converter would require major truncation of the input word to a 16-bit length, thus losing the additional accuracy provided by the remaining 8 bits. As for quantization error, the addition of 2 bits to the 16-bit word provides 4 times as many quantizable voltage levels, thus allowing for more exact measurement of the analog voltage when converted into digital signals.
One solution in the prior art is to deal with the smaller digital word size which possesses a lower dynamic range was to utilize a process known as dithering. Dithering eliminates harmonic distortion caused by quantization. However, the process tends to create a higher SNR. There are 2 widely used methods for producing the dithering effect, Broadband-Triangular dither and Weighted dither.
Accordingly it is an object of the present invention to be able to process digital audio signal data produced by sources using the 16-bit format as well as to be able to accept up to 24-bit digital audio words.
Another object of the present invention is to provide for the automatic adjustment for input word lengths greater than 16-bits and to properly dither to 18-bits.
Another object of the present invention is to provide for hybrid digital/analog volume control. In the prior art, a digital output signal could be attenuated in either of 2 ways, either in the digital domain before the digital-to-analog conversion or in the analog domain after the digital-to-analog conversion. There are advantages and disadvantages to each of these approaches. Implementing attenuation in the digital domain is done by multiplying the digital word by a gain value which produces extra precision bits in the digital audio word which is then dithered and truncated to the word size of the digital-to-analog converter. This results in a constant noise floor during attenuation, thus as the audio signal becomes smaller, it begins to phase into the constant noise, thus making digital attenuation less optimal in low volume situations.
Implementing attenuation in the analog domain is done with the use of mechanical potentiometers which are often the source of mechanical problems and mechanical "wiper noise" which may be most disturbing at higher gain settings. In addition, analog attenuation often requires the use of dual volume potentiometers which require balancing; and the necessity for connections between the potentiometer and the analog board which may introduce additional noise into the audio output. Analog gain control is however potentially free of constant background noise that is inherent in the digital implementation.
Accordingly the present invention implements a hybrid attenuation scheme where by attenuation in the "lower" part of the gain range is performed digitally and attenuation in the "upper" part of the gain range is performed by analog means.
It is yet another object of the present invention to provide both Broadband-Triangular and Weighted dithering algorithms for user selection to analyze digital audio data of word length greater than 18 bits and when volume control is below the digital attenuation threshold.
These and other objects of the invention will become apparent in light of the present specification and drawings.